Estimated reading time: 2 minutes

Vias, Modeling, and Signal Integrity

Editor Andy Shaughnessy recently sprung a request on me: Would I like to write an item on vias? A complex and tricky subject, I thought. But after some reflection it seemed like a good idea…so, here goes.

I was just about to leave for PCB West in September, and my connecting flight developed a fault. Always better when the plane breaks on the ground than in the air, but inconvenient nevertheless. If you want to be certain to catch your long-haul flight, says the ground agent, you must go via London Gatwick.

I’ll cut a long story short, but I had 10 minutes to decide. Should I, on the one hand, get to London Heathrow via London Gatwick, or take my original connection which maybe will, maybe won’t, get me to Heathrow in time for my connection? I bit the bullet, had my bags offloaded and started the process of looping through security and booking a new flight on an alternate airline only 45 minutes before departure whilst at the same time trying to book an overnight hotel whilst going through security to the new flight.

In the process, I became one of those annoying people who is on the phone at the gate and up the jetway. I did try to book online but that’s a mare too when you are in a rush. Still, I got the last room available close to the desired terminal.

What’s all that got to do signal integrity? Well, it’s a lot easier to go somewhere by a direct route than via somewhere else. The same is true for signals. The best a designer can do is to ensure the via offers the least inconvenience to the signal; otherwise it’s going to end up at its destination a bit grumpy and battered out of shape, just like I was when I boarded the aircraft for the long-haul flight the following morning. Though a few glasses of shiraz later, my mental state had equalised. But signals aren’t always so lucky.

Before delving down into the electrical via, I would like you to have a think about the following scenario on a transmission line. Bear with me: Understanding the following will help make more sense of the via scenario. The second thing I would like you to bear in mind simultaneously is that good modeling can’t fix a bad design. The model can tell you where a design is weak, but if you have committed your design to product, the model can only tell you how it behaves. Some less experienced designers seem to fall into the trap of thinking a better model will fix something that doesn’t work; it won’t. It will only reassure you that the design was bad in the first place.

To read this entire article, which appeared in the November 2016 issue, click here.